Artificial Intelligence

2007 Submissions

[12] viXra:2007.0209 [pdf] submitted on 2020-07-27 06:26:13

Lunar Rock Classification Using Machine Learning

Authors: Arshita Kalra, Arnav Bhavsar
Comments: 5 Pages.

Lunar landings by esteemed space stations around the world have yielded an abundance of new scientific data on the Moon which has helped scientists to study our closest neighbour and hence have provided evidence for understanding Earth’s past and future. This paper is about solving the challenge on HackerEarth about classifying the lunar rock into small or large rock. These tasks have historically been conducted by visual image inspection, thereby reducing the scope, reliability and accuracy of the retrieval. The competition was to build a machine learning model to reduce human effort of doing a monotonous task. We built a Support Vector Machine model, used widely in classification problems, feeding features extracted from images in the dataset using OpenCV, only to obtain an accuracy of 99.41%. Our source code solving the challenge and the dataset are given in the github repository https://github.com/ArshitaKalra/Lunar-Rock-classification.
Category: Artificial Intelligence

[11] viXra:2007.0200 [pdf] submitted on 2020-07-24 19:27:40

A Semantic Question Answering in a Restricted Smart Factory Domain Attaching to Various Data Sources

Authors: Orçun Oruç
Comments: 24 Pages.

Industrial manufacturing has become more interconnected between smart devices such as the industry of things edge devices, tablets, manufacturing equipment, and smartphones. Smart factories have emerged and evolved with digital technologies and data science in manufacturing systems over the past few years. Smart factories make complex data enables digital manufacturing and smart supply chain management and enhanced assembly line control. Nowadays, smart factories produce a large amount of data that needs to be apprehensible by human operators and experts in decision making. However, linked data is still hard to understand and interpret for human operators, thus we need a translating system from linked data to natural language or summarization of the volume of linked data by eliminating undesired results in the linked data repository. In this study, we propose a semantic question answering in a restricted smart factory domain attaching to various data sources. In the end, we will perform qualitative and quantitative evaluation of the semantic question answering, as well as discuss findings and conclude the main points with regard to our research questions.
Category: Artificial Intelligence

[9] viXra:2007.0085 [pdf] replaced on 2020-12-29 20:10:57

Microscopy Image Processing for the Human Eye

Authors: Zeyue Xia, Mohamad Nadim Barakat, Serri Matula, Zijun Hui, John Stavrakakis
Comments: 7 Pages. Computer Vision

Vivo confocal microscopy allows scientists to better understand eye health and systemic diseases. Microneuromas could play a role, however, monitoring their growth from a mosaic of images is error prone and time consuming. We used automated image stitching as a solution; focusing on accuracy and computational speed of three different feature detection algorithms: SIFT, SURF, and ORB. The results illustrated that SURF was computationally efficient with our data. Future investigation is to create a global solution that can replace the need for manual image stitching in this application.
Category: Artificial Intelligence

[8] viXra:2007.0084 [pdf] submitted on 2020-07-12 21:45:42

Nonextensive Belief Entropy

Authors: Yige Xue, Yong Deng
Comments: 16 Pages.

The belief entropy has high performance in handling uncertain information, which is the extension of information entropy in Dempster-shafer evidence theory. The Tsallis entropy is an extent of information entropy, which is a nonextensive entropy. However, how to applied the idea of belief entropy to improve the Tsallis entropy is still an open issue. This paper proposes the nonextensive belief entropy(NBE), which consists of belief entropy and Tsallis entropy. If the extensive constant of the proposed model equal to 1, then the NBE will degenerate into classical belief entropy. Furthermore, When the basic probability assignment degenerates into probability distribution, then the proposed entropy will be degenerated as classical Tsallis entropy. Meanwhile, if NBE focus on the probability distribution and the extensive constant equal to 1, then the NBE is equate the classical information entropy. Numerical examples are applied to prove the efficiency of the proposed entropy. The experimental results show that the proposed entropy can combine the belief entropy and Tsallis entropy effectively and successfully.
Category: Artificial Intelligence

[7] viXra:2007.0040 [pdf] submitted on 2020-07-06 11:54:41

Paper Car Speed Detection Using Computer Vision

Authors: Aditi Singh, Raju Ranjan
Comments: 3 Pages.

When it comes to road safety, detection and monitoring of car speed is one of the major tasks. The use of a simple camera and image processing software eliminated the primary tools of speed detection like handheld radar gun. In these techniques, the speed is calculated as the car passes through the camera’s field of view (FOV). The speed is calculated by noting the time taken by car between entering and exiting FOV. Some systems used individual cameras at entry and exit FOVs. Thus, it does now calculate the speed in between this interval. This paper proposes a technique to measure speed of car the moment it enters into the camera’s FOV till the time it exits the FOV. Using the Deep Learning Single Shot Detector (SSD) implemented using Convolutional Neural Network (CNN), the cars entering FOV are detected and based on the distance they travel in FOV and time taken to cover that distance the speed of car is calculated
Category: Artificial Intelligence

[6] viXra:2007.0039 [pdf] submitted on 2020-07-06 20:06:00

Eye Gaze Optical Mouse using Computer Vision

Authors: Dhananjay Mewati, Jerald Nirmal Kumar
Comments: 3 Pages.

This paper proposes a technique of using the movement of eyes to control the movement of cursor on monitor screens. Thereby, creating new ways of Human Computer Interaction (HCI) and also helping physically handicapped people to interact with computer devices more efficiently. Earlier eye gaze optical mouse comprised of a head gear which had an eye motion sensor attached and were more hardware based. The input gathered through these sensors helped in cursor movement on screen. With the advancement in the field of Image Processing Techniques and Artificial Intelligence, a simple web camera attached with computer can be used to perform this task. In this paper, pupil of the eye is detected. The coordinates gathered by tracking pupil movement are mapped with the coordinate of display monitor. Based on this mapping the mouse cursor can be moved on the screen.
Category: Artificial Intelligence

[5] viXra:2007.0034 [pdf] submitted on 2020-07-05 21:02:19

Dynamic and System Agnostic Malware Detection Via Machine Learning

Authors: Michael Sgroi, Doug Jacobson
Comments: 23 Pages.

This paper discusses malware detection in personal computers. Current malware detection solutions are static. Antiviruses rely on lists of malicious signatures that are then used in file scanning. These antiviruses are also very dependent on the operating system, requiring different solutions for different systems. This paper presents a solution that detects malware based on runtime attributes. It also emphasizes that these attributes are easily accessible and fairly generic meaning that it functions across systems and without specialized information. The attributes are used in a machine learning system that makes it flexible for retraining if necessary, but capable of handling new variants without needing to modify the solution. It can also be run quickly which allows for detection to be achieved before the malware gets too far.
Category: Artificial Intelligence

[4] viXra:2007.0033 [pdf] submitted on 2020-07-05 21:21:29

Sentiment Analysis on Tweets Using Document Embeddings

Authors: Qasim Nawaz
Comments: 33 Pages. N/A

Sentiment Analysis is one of the primary areas of natural language processing and information retrieval being tackled by researchers to date, and for good reason; the internet. The internet is a mostly untapped source of rich amounts of data that can be used to gauge the opinions of people, in reference to any number of topics. Twitter is one such platform designed for people to voice their opinions in the form of tweets about any topic they desire. My project will set out to investigate the best way to be able to analyse the sentiment of these aforementioned tweets using machine learning techniques. I will be training word vector-based, and paragraph vector-based models on a dataset consisting of 1.6 million tweets, in conjunction with various classifiers in order to find the best performing method in which to obtain the sentiment of tweets.
Category: Artificial Intelligence

[3] viXra:2007.0031 [pdf] submitted on 2020-07-06 04:09:56

Neural Network based Classification of Flowers using Transfer Learning

Authors: Abhishek
Comments: 3 Pages.

One of the classical problems in the field of computer vision and machine learning and subsequently deep learning is image classification. While Deep Learning solves the much difficult hurdles like feature extraction and presents us with better optimizations like gradient descent and Adam optimizer, most deep learning models still need a lot of raw computational power to train models on local Graphical Processing Units (GPUs) or Tensor Processing Units (TPUs) in the cloud. All of this computational power is not readily available in all environments and systems and hence the concept of pre-trained models can help to reduce training time by a huge margin. Initial models get trained on large array of GPUs and do feature extraction. The classification part is for the end-user to customize in accordance to the problem at hand and can be completed in very less time. We tackled the multi-class classification botanical problem of identifying flowers of 5 types, namely, Sunflower, Rose, Dandelion, Daisy, and Tulip. The feature extraction part is done with the model (Google’s Inception-v3) and fully connected softmax layers were trained on local machine on a Nvidia GeForce GTX 950 (with CUDA activated) within 30 minutes time and total steps/epochs were 4000 only. The total number of training images is 3,500 (approx.). The finished model produced results with final test accuracy as 91.9% on new images (N=664).
Category: Artificial Intelligence

[2] viXra:2007.0030 [pdf] submitted on 2020-07-06 04:28:45

Automatic Text SummarizerApplication Using Extractive Text Summarization

Authors: Ritesh Kumar Bharadwaj
Comments: 5 Pages.

Text Summarization as a phenomenon has always been present and rather an evolving one with the advent of new technologies both in terms of data collection as well for the processing of this data. One reason of using text summarization is the huge amount of data floating over the internet in the form of text files, comments which is though potent enough to be used to extract useful information. but since the amount of text present in these sources is too huge, so the need of text summarization becomes justified by every argument. Some of the areas where text summarization is vastly used is applications involved in providing capsule information such as compact news applications, or websites providing academic notes for various examinations This paper presents an auto text summarizer application which takes the URL of a web page as input, performs summarization on the selected elements and then presents this summarized text content on the front end of a web application. At the backend, the process of scraping of web page content (if an http URL is provided as input) using beautiful soup library or reading of text provided takes place. news in short forms, or micro blogging websites. The scraped content after being preprocessed properly is summarized using a suitable library which in our case is one among NLTK, Spacy, Genism and Sumy. The summarized content is presented at the frontend using flask framework of Python. The results produced using different libraries are compared in the end in terms of reading time of the summarized content. The application uses extractive text summarization technique in order to achieve its result which is a compact summary of the textual data prepared from the keywords already present in the document Keywords: Auto Text Summarizer, URL, Flask, Web Scraping, Nltk, Spacy, Sumy, Gensim, Extractive Text Summarization
Category: Artificial Intelligence

[1] viXra:2007.0029 [pdf] submitted on 2020-07-06 05:18:45

Lung Cancer Detection using CNN

Authors: Mohammed Tahir
Comments: 3 Pages.

The recent surge of Deep Learning has led to breakthrough advancements in almost every field of its application. A particular deep learning architecture, arguably the most popular one is the Convolution Neural Networks. The interest in convnets has seen an exponential increase due to their effectiveness and scalability. CNNs have become the go-to solution for image data problems and has provided results that are at par with if not better than human standards. The simplicity of the CNN architecture is another big factor of its success. The image processing and classification capabilities of CNN have found great usage in medical field, making it possible to detect and classify diseases as severe as Cancer effectively for the sake of better care. In this project, I’ve initiated an elaborate study of Convolution Neural Networks, built multiple architectures from scratch and furthered our understanding with the preparation of an elementary dog-cat CNN classifier model followed by a more extensive CNN model for detection of lung cancer in a patient. The project is built on Google’s interactive and versatile cloud platform for AI development Google Colaboratory, using the open-source neural network library ‘Keras’ for model development and libraries such as matplotlib and tensorboard (tensorflow) for result plotting and analysis. Data for training and testing our model was extracted from the ‘ LUNA 2016 medical image database ’. The model was tuned using Grid-Search and achieved over 97% test accuracy in its final iterations. To culminate,I have enlisted some future-work prospects like De-convolution/Translated-Convolution,implement one or more named CNN networks like Inception or Alexnet, test the model on larger images etc
Category: Artificial Intelligence